Quantifying the Accuracy of Density Functionals on Transition Metal Bulk and Surface Properties

Area of Science:

• Computational materials science
• Solid-state physics
• Quantum chemistry

Background:

• Density functional theory (DFT) relies on accurate exchange-correlation (xc) functionals, which are not known exactly.
• Generalized gradient approximation (GGA) functionals like PBE and VV have shown promise for describing transition metal (TM) bulk and surface properties.
• Previous assessments omitted several xc functionals, including local density approximation (LDA) and some GGAs, as well as newer meta-GGAs and machine learning-based functionals.

Purpose of the Study:

• To systematically assess a wider range of exchange-correlation (xc) functionals for transition metal (TM) bulk and surface properties.
• To compare the performance of newly included functionals, such as AM05, SCAN, and BEEF, against established functionals like PBE and VV.
• To identify the most suitable xc functionals for accurate predictions of TM properties.

Main Methods:

• Utilized density functional theory (DFT) calculations to evaluate various exchange-correlation (xc) functionals.
• Assessed bulk properties including interatomic bond distance, cohesive energy, and bulk modulus.
• Evaluated surface properties such as surface energy, work function, and interlayer distances.

Main Results:

• No newly assessed xc functionals outperformed PBE or VV for transition metal properties.
• The Armiento-Mattsson (AM05) and Strongly Constrained Appropriately Normed (SCAN) functionals demonstrated acceptable performance.
• The Bayesian Error Estimation Functional (BEEF) showed potential but requires further parametrization for competitive accuracy.

Conclusions:

• Existing GGA functionals like PBE and VV remain the most reliable for transition metal systems.
• AM05 and SCAN represent viable alternatives, offering acceptable accuracy for TM property predictions.
• Further development of machine learning-based functionals like BEEF is needed to enhance their predictive power.